PolicyIK overview showing a robot arm reaching a target pose, precision error curves, and results from learning directly on hardware.

PolicyIK

High-Precision Inverse Kinematics for Robot Arms with Reinforcement Learning

Yuhan Liu*, Xinqi Liu*, Zixuan Dong*, Keith W. Ross, Shenji Wan
* Equal contribution

A goal-conditioned reinforcement learning approach to high-precision inverse kinematics, enabling robot arms to reach and hold target poses in simulation and on physical hardware, with applications to vision-language-action manipulation.

PanoTac tactile sensors mounted on a robot gripper, alongside an exploded diagram of the sensor's layered structure.

PanoTac

A Dense Three-Axis Tactile Sensor for Robust Touch-Centric Manipulation

Binglin (Andre) Wang, Xiaole Jiang, Zicheng Teng, Xinqi Liu, Yunzhu Li, Lerrel Pinto, Zichen Jeff Cui, Binghao Huang

A low-cost, compliant tactile fingertip combining dense normal and shear feedback for touch-centric robot manipulation, including test tube preparation, grain scooping, and salt pouring.

Figure 1 from the RUKA-v2 paper: the real robotic hand grasping a block, with its parallel wrist, side mount, magnetic encoders, and soft fingertips highlighted.

Accepted · IEEE Humanoids 2026

RUKA-v2

Tendon Driven Open-Source Dexterous Hand with Wrist and Abduction for Robot Learning

Xinqi Liu, Ruoxi Hu, Alejandro Ojeda Olarte, Zhuoran Chen, Kenny Ma, Charles Cheng Ji, Lerrel Pinto, Raunaq Bhirangi, Irmak Guzey

ICRA Workshop · Spotlight

An open-source, tendon-driven humanoid hand with a decoupled 2-DOF wrist and finger abduction/adduction for dexterous teleoperation and robot learning.

More projects and publications coming soon.